The Role of the Bone Marrow Immune Niche in Preventing Relapse in Adult B-ALL Following Reduced Intensity Conditioning Allogeneic HSCT

Reduced intensity-conditioned allogeneic haematopoietic stem cell transplants (alloHSCT) improved event-free survival for patients with B-cell acute lymphoblastic leukaemia (B-ALL) >40 years old, as established by the UKALL14 trial (Marks et al. 2022) The main factor predicting outcome post alloHSCT was persistence of minimal residual disease (MRD). Other studies also find that MRD to be a strong predictor of relapse after alloHSCT in ALL (Della Starza et al., 2019) which is associated with an ineffective graft-versus-leukaemia (GVL) response by donor T cells (Rosko et al., 2016); (Lin et al., 2000).

We hypothesised that we would find a different T cell ‘signature’ in patients who relapsed and those who did not following alloHSCT in the UKALL14 study. We used post-transplant samples to assess the bone marrow (BM) immune cell content. Twenty-one BM samples from 12 patients (aged 41-60, median = 48 years old) were analysed by bulk RNA-sequencing (RNA-seq), ranging from 8-109 weeks post-alloHSCT. Six (50%) patients were female and six (50%) male, 2 (17%) had BCR::ABL+ B-ALL, and 2 (17%) had other UKALL14 high-risk cytogenetics. Samples were analysed in 2 groups: A) patients whose disease was in continuous remission and remained in remission long term (CR → CR); B) patients whose disease was in complete remission but relapsed in the future (CR → Rel). When comparing 1 sample per patient (13-55 weeks post-alloHSCT, median = 38 weeks), we did not detect any differences in the T cell subpopulation representation. However, neutrophil-expressed genes were significantly upregulated (L2FC ≥ 1, p-value ≤ 0.05), and gene set enrichment analysis (GSEA) revealed an enrichment of pathways associated with neutrophil degranulation in CR → Rel, compared to CR → CR. Longitudinal sampling, used to assess how the BM composition changes temporally, post-alloHSCT, did not identify any significant differences across the cohort over time.

To characterise this myeloid signature at greater resolution, single cell RNA-sequencing was performed on BM samples from 10 patients (aged 36-59, median = 44.5 years old), in the same analysis groups as previously described, with samples taken between 19-82 weeks post-alloHSCT (median = 31 weeks). Seven (58%) were male and 5 (42%) female, 3 had BCR::ABL+ B-ALL, and an additional 5 (42%) had high-risk cytogenetics. Investigation of this cohort revealed increased cytotoxic CD8⁺ T cells and NK cells in CR → CR, with CD4⁺ naïve and central memory populations more prominent in CR → Rel. A neutrophil population with a gene expression signature comparable to that seen in the bulk RNA-seq was also enriched in CR → Rel. Cell-cell interactome analysis revealed stronger putative interactions in CR → Rel specimens, most notably involving the CD4⁺ naïve T cell and neutrophil populations.

To determine whether this myeloid signature was transplant-dependent, we analysed data from bulk RNA-seq of 7 BM diagnostic samples, from patients with samples analysed post-alloHSCT, for the presence of the same signature. Analysis using DESeq2 and GSEA identified a myeloid gene signature that was associated with subsequent relapse after alloHSCT; this gene signature overlapped with the myeloid signature found in post-transplant CR → Rel (hypergeometric test p = 8.5x10-36). Six myeloid signatures were generated from the post-alloHSCT and diagnostic bulk RNA-seq and used to generate normalised enrichment scores (NES) for a cohort of 150 UKALL14 patients at diagnosis. Interrogation of this cohort showed that increased NES of all six signatures were significantly correlated with age and event-free survival.

To summarise, we have identified BM immune signatures associated with continuous remission versus future relapse after alloHSCT for B-ALL. Our data suggest that a neutrophil signature is enriched in both the diagnostic and post-transplant samples of patients destined to relapse, which is supported by its correlation with age and EFS in a large diagnostic cohort. To corroborate whether this myeloid population predicts eventual treatment failure post-alloHSCT in an independent cohort, the same 6 signatures will be used to interrogate bulk RNA-seq of patients at diagnosis from the GRALL trial. Further work is being performed to evaluate the profile of the neutrophil-like populations in the bone marrow samples of relapsing patients and to determine how they might regulate anti-tumour immune surveillance.